[Info-vax] Are errors on PEA0: dangerous?

[Hans Bachner]

Phillip Helbig---undress to reply <helbig at astro.multiCLOTHESvax.de> 
wrote:

> As I recently mentioned here, after using LANCP to move from 10 Mb/s
> HD to 100 Mb/s FD on two PWS in my cluster, the slowly rising (few
> errors per day) error count on PEA0: (and the somewhat lower number on
> the SCSI disks) on an XP1000 in the cluster, which has always had 100
> Mb/s and FD, stopped rising, completely.  However, since then PEA0: on
> the two PWS have been rising slowly as well (perhaps a bit more
> quickly than on the XP1000).
> 
> How dangerous are errors on PEA0:?

depends on the kind of error - but if an error is recognized, it 
probably will be handled.

If OpenVMS detects errors critical to system integrity, it usually 
bugchecks.

> Any idea what causes this?

What does $ ANALYZE /ERROR [/ELV] tell you about the cause?

> Any idea how to fix it?

Without knowing the details, I suspect that the switch starts auto-
negotiating or tries different speed/duplex settings from time to time 
depending on events on the link (delays, link level transmission errors 
etc.)

I'd go for a managed switch and set the port to 100/FD, as well as 
EWA0_MODE at the console. 

There are (relatively) low cost 8-port manageable switches available, 
e.g. from HP or Cisco.

> Is there any setting I might need to change at the console level which 
> can't be changed from within LANCP?
> 
> Since the change, I also notice, when logged in over LAT, no terminal 
> (emulator) output for a few seconds.  This doesn't happen often, but
> has been happening since the change.

Do these events correlate to PEA0 error count increases or entries in 
ERRLOG.SYS?

Hans.